Eom refs.pl

#!/usr/bin/perl -w

# This script operates like a filter, i.e, it reads a file from
# standard input, rewrites its content and writes the transcript to
# standard output.

# It rewrites old eom reference tables into wikipedia table style. It uses
# the Ref and Cite templates.

# It replaces the reference descriptors (like [1], [2],
# etc. by the first two letters of the author names like [Ab], [Ha],
# [KuZi] etc.  and orders the tables alphabetically with respect to
# these descriptors.  Moreover, in the tables, anchors are set by
# {{Ref|Ab}} using the Ref template while, within the text, references
# {{Cite|Ab}} to these anchors are installed by the Cite template.

# Author names are retrieved by the following heuristic: 

# All those words after the ref descriptor (in brackets) are collected
# wich start with a capital followed by a small letter and ended by a
# comma (like Abel, ), and the first two letters of all these words
# are concatenated and used as new reference descriptor [Ab]. 

# In case of multiplicities of such strings the second, third, ... gets a
# '2','3', ... appended like [Ab2], [Ab3] etc.


use strict;
use utf8;
use Encode;
use  Unicode::Collate;

# get file content:
undef $/;
my $f = decode('UTF-8', readline STDIN);




$f = &transcribe($f);
$f = encode('UTF-8',$f);
#print "...............\n";
print "{{MSC|}}\n{{TEX|done}}\n\n";
print $f;
exit(0);

sub transcribe {
    my $f = $_[0];
# for unicode sorting/collating, see below
# Package libunicode-collate-perl required
    my $Collator = Unicode::Collate->new();

# collect all reference lists in array 
    my @refs = ($f =~ /\=+References\=+\s+(\<table\>.*?\<\/table\>)/sg); 
    if ($#refs < 0) { 
	my $error = <<"END_ERR";
Error: no refs: length of refs array: $#refs.

There is no References section in the old EoM style, using the format
between the html tags <table> ... </table>.

Reasons for this could be that the references have been already
processed into new (Mediawiki) style, or that they are just scrambled,
or that there are no references at all in the article.

Please check or contact one of the editors.
END_ERR

        print $error;
	exit(1);
    }

    my %H = (); # Hash for ref keys like [1] etc.
    my %K = (); # Hash for multiplicity of ref key occurence

# Array for all initial keys:
    my @items = ($f =~ /\>\[(\w+?)\]\<.*?,/sg);
    foreach(@items) { 
	my $key = $_;
	my $y = "";
	# find names after $key and before next comma:
	if ($f =~ /\>\[$key\]\<(.*?),\s*["\']/sg) {
	    my $x = $1;
	    # remove material in [..] (alternative name spelling)
	    $x =~ s/\[.*?\]//sg;
	    # collect first two letters of names 
	    # and concat into $y, will serve as new key:
#	    my @A = ($x =~ / ([A-Z]\w)[^\.]/sg);
            # required for unicode match:
	    my @A = ($x =~ / (\p{Lu}\p{Ll})[^\.]/sg);
	    foreach(@A) { $y .= $_;	}
	}
	$K{$y}++; 
	if ($K{$y} > 1) { # append count if bigger than 1
	    $H{$key} = $y.$K{$y};
	} else {
	    $H{$key} = $y;
	}
    }

    $f =~ s/\[\[#References\|\[(\w+?)\]\]\]/{{Cite|$H{$1}}}/sg;
    $f =~ s/\|\s*\[(\w?)\]\s*\|/|{{Ref|$1}}|/sg;


    foreach(@refs) { # process all ref lists:
#    print "$_\n";
	my $x = $_;
	# do table conversion:
	my $y = &replace($x);
	foreach(keys %H) {
	    $y =~ s/\[$_\]/{{Ref|$H{$_}}}/sg;
	}
	# sort bib entries by ref keys 
	my %R = ();
	my @A = split(/\|\-/,$y);
	foreach(@A) {
	    if (/\{\{Ref\|(\w+)\}\}/) { $R{$1} = $_; }
	}
	my $z = '{|'."\n".'|-';
#	foreach(sort keys %R) { $z .= $R{$_}.'|-';  }
	foreach($Collator->sort(keys %R)) { $z .= $R{$_}.'|-';  }

	$z .= "\n\|\}\n";
	
	# do replacements:
	# prepend a backslash before each of <>"=()[]+-?|^$*\~
	$x =~ s/([\<\>\"\=\(\)\[\]\+\-\?\|\^\$\*\\\~])/\\$1/sg;
	$f =~ s/$x/$z/sg;
    }
    return $f;
}

## converts bib table from html to wiki structure
sub replace{ 

## cf. http://en.wikipedia.org/wiki/Help:Table#Pipe_syntax_tutorial
    my $f = $_[0];
    $f =~ s/\n/ /sg;
    # replace data cells (last | still to be removed!):
    $f =~ s/\<td\s(.*?)\>(.*?)\<\/td\>/|$1|$2|/sgi;
    # replace row (consider removal of last | from above):
    $f =~ s/\<tr\>(.*?)[\s\|]*\<\/tr\>\s*/|\-\n$1\n/sgi;
    # no spaces between consecutive |:
    $f =~s/\|\s*\|/||/sgi;
    # replace <table> and </table>:
    $f =~ s/\<table\>/\{|\n/sgi;
    $f =~ s/\<\/table\>/|\-\n|\}/sgi;
    # remove possible | before template calls:
    $f =~ s/\|\s*\{\{/\{\{/sg;
    # remove space before some punctuation:
    $f =~ s/ +([\,\.])/$1/sg;
    return $f;
}